In partnership with

Get more meetings by removing the LinkedIn limits.

When you need more leads, you want to do more, right?

  • Adding mailboxes to cold emails.

  • Placing more cold calls

  • Running more ads.

But what do you do if your leads are on LinkedIn? LinkedIn outreach still depends on the limited number of profiles already on your team.

Aimfox Avatars lets you rent dedicated, customizable LinkedIn profiles that start conversations and hand interested replies to your existing sellers.

Add outreach capacity without hiring another SDR or turning every employee account into a prospecting channel..

Beginners in AI

Good morning and thank you for joining us again!

Welcome to this daily edition of Beginners in AI, where we explore the latest trends, tools, and news in the world of AI and the tech that surrounds it. Like all editions, this is human curated and edited, and published with the intention of making AI news and technology more accessible to everyone.

THE FRONT PAGE

Google's Voice AI Now Works While It Talks, and It's in Gmail Today

TLDR: Google released two voice models, Gemini 3.8 Live and 3.8 Live Extended Thinking, that see through your camera, switch languages on their own, and finish jobs in the background while you keep talking.

The Story:
Both models went out September 15. It reads video as you go, so you can point your phone at a jammed printer and ask what the blinking light means. It also detects and switches between 97 languages without being told. Both can run tools in the background and keep the conversation going while the work finishes. Extended Thinking narrates long jobs while it does them, dropping in a "Let me check that" so you're not sitting in silence wondering if it died.

Its Significance:
Voice assistants have felt clumsy for a reason: dead air. Ask for something that takes twelve seconds and you get nothing until it's done, which is why most people give up and go back to typing. Speaking and working at the same time is closer to how a decent coworker behaves, and Google put it straight into Gmail, Docs, Keep, and Search Live rather than a demo page. Voice is the one way to use a computer when your hands are full and your eyes are somewhere else and will likely become the dominant way that we interact with AI models in the future.

1-on-1 Live Claude AI Crash Course Tutorial
1-on-1 Live Claude AI Crash Course Tutorial
A 1-hour beginner-friendly video call to get you comfortable with the Claude ecosystem — Claude Desktop, Claude Code, Cowork, Skills, Projects, Correct File set-ups, and Plugins. Real-world example...
$75.00 usd

QUICK TAKES

The story: Jennifer Huddleston of the Cato Institute wrote Monday that a government-ordered pause on AI development would likely fail at safety while shielding the biggest labs from competition, since rules written around today's models can lock in whoever already has the money to follow them. Block co-founder Jack Dorsey posted the same day that he wants independent testing and published limitations, but opposes industry-wide caps worked out by the current leaders, writing that protecting a company's commercial advantage isn't a safety goal.

Your takeaway: You've seen this in banking, healthcare, and housing. Big firms keep lawyers and compliance staff on payroll for whatever rules show up, and a four-person startup can't, so the paperwork becomes the wall. Dorsey does leave one door open: holding back a model when there's real evidence it would raise catastrophic risk, which is a much narrower test than a blanket pause.

The story: Google added live voice conversations with your own notes in nearly 100 languages, an audio recorder in the mobile app for lectures, and quizzes and flashcards built from whatever you upload. College students in the US can claim a year of Google AI Pro free, a plan that normally runs $19.99 a month.

Your takeaway: Every answer stays tied to the sources you loaded, so it can't wander off inventing facts the way an open chatbot will. That's the part worth stealing even if you left school decades ago. Dump a stack of work PDFs in and talk your way through them on the drive home.

The story: Reuters reported the two companies are discussing a deal that would put Korean memory production on US soil for the first time, either by leasing part of Intel's Ohio plant or through a joint venture with cloud providers. SK Hynix told TechCrunch that nothing is final and no decision has been made on either option.

Your takeaway: We need this one. Tech companies have swallowed so much memory that prices are skyrocketing for regular buyers, and Intel's Ohio site, announced in 2022 with up to $100 billion behind it, has slipped to 2030 and 2031. Putting chips in an empty building helps the shortage and a struggling chipmaker the US government part-owns.

TOOLS ON OUR RADAR

🎬 Descript Freemium: Edit video and podcasts by editing a text transcript instead of a timeline, the free plan includes limited transcription minutes and adds a watermark on exports.

📊 Umami Free and Open Source: Track website visitors without cookies, the Cloud Hobby tier is free forever for 100k events a month, self-hosting is also always free.

🤝 Copper CRM Paid: Manage contacts and deals inside Gmail and Google Workspace without switching tabs, no ongoing free plan, just a 14-day trial before subscription pricing kicks in.

🧑‍💻 Open Cowork Free and Open Source: An AI agent that runs on your own desktop and can browse, edit files, and generate documents for you, one-click installer, no coding required.

TRENDING

Reddit co-founder Alexis Ohanian says the industry has explained AI badly, telling CNBC that tech has done a "pretty tone deaf job" of it. He argued the real dangers are more ordinary than Terminator scenarios, pointing to groups of AI agents working together, and compared AI to nuclear energy: powerful, harmful if handled badly, useful if managed well.

Google published a tally of what its AI is doing for science, and some of the numbers are worth knowing. A breast cancer study with Imperial College London and the UK's NHS found AI caught 25% of the cancers missed between screenings across mammograms from 175,000 women. Flood forecasts now cover 2 billion people in more than 150 countries.

Meta bundled everything into one subscription called Meta One, pulling paid features across Instagram, Facebook, WhatsApp and Meta AI into a single plan with 50-plus features. Prices start at $2.99 a month for one app and top out at $499 for the largest business tier. Meta says the free versions aren't changing and that 15 million subscriptions and trials are signed up so far.

A startup built an AI model that talks to software instead of people. TypeSafe AI, funded with $40 million, released Jev, which returns typed decisions rather than sentences. The demo has it playing Doom at roughly 10 decisions per second for about $7 an hour, reading a text description of the game rather than the picture on screen.

Google DeepMind tested whether AI knows when to keep quiet, running GPT-4o, Gemma 3 27B, DeepSeek-V3 and Qwen3-Next through a four-phase experiment published in Nature Machine Intelligence. The models skipped questions when their internal confidence was low, and dialing that confidence up or down inside Gemma changed how often it passed. Useful, since models are being handed jobs where a confident wrong answer costs real money.

Meta now lets an AI agent set up WhatsApp Business for you. Creating the account, verifying the phone number, registering for API access, checking terms of service, all of it can be handed to a coding assistant through a new tool Meta opened up. That work used to mean bouncing between four separate consoles.

TRY THIS PROMPT (copy and paste into Claude, Grok, ChatGPT, or Gemini)

📻 Enter a year. Get what was actually #1 on the radio, in theaters, and in the news, not a guess dressed up as trivia.

Build a single-file HTML app with vanilla HTML/CSS/JS. The Nostalgia Time Capsule  enter a year (and optional date), get real documented facts via live web search. No API key needed.

Aesthetic: warm dark brown (#1e1610), retro orange (#e07a3c) primary with a glow top-left, retro teal (#4ca8a0) secondary for cost stats with a glow bottom-right. Abril Fatface (bold retro display) for the giant year number, Inter for body, JetBrains Mono for labels.

Form: birth year number input, optional exact-date text input.

Technical: call the Anthropic Messages API WITH tools:[{type:'web_search_20250305', name:'web_search'}] (works automatically), system prompt as a nostalgia researcher who finds real, well-documented facts for the given year  #1 song (title/artist only, never lyrics), top box-office movie, a real major event, 2-4 real approximate period prices  and, if a date was given, a real event from that specific date, never inventing anything not found via search. Return raw JSON: year, song_title_artist, top_movie, notable_event (2-3 sentences), cost_snapshot (3-4: item + price), fun_fact, exact_date_note (empty if no date given or nothing found).

Render: a huge retro display-font year number. Two highlight cards (song, movie). A teal-bordered notable-event card. A cost-snapshot grid. An italic fun-fact card. An optional date-specific card when a date was provided.

What this does: Type a birth year, or any year, plus an optional exact date, and it searches the live web for real documented facts: the genuine #1 song of that year, the actual top movie, a real major event, and a handful of real approximate prices from the time, gas, a house, a movie ticket, so you get a tangible sense of the era.

What this looks like:

Your traffic is fine. Your signups aren't.

Visitors land and leave, and "looks fine to me" isn't a diagnosis. SureThing audits SEO, speed, mobile, and messaging against the page, then ranks the fixes by impact.

WHERE WE STAND(based on today’s news)

AI Can Now: Return a decision in 70 to 500 milliseconds that always matches the exact format the software asked for, picking from as many as 255 options.

Still Can't: Be right. TypeSafe says the zero on its hallucination chart means the shape always matches, not that the choice is correct.

AI Can Now: Hold back an answer when its own internal confidence falls below a line, and adjust that line when you tell it to.

Still Can't: Report that confidence cleanly. The DeepMind paper found both the spoken confidence and the number scores are rough copies of something richer going on inside.

FROM THE WEB

RECOMMENDED LISTENING/READING/WATCHING

Zone 414 (2021) - Movie

Guy Pearce plays a burned-out private investigator hired to search for a missing woman inside a walled colony of advanced androids, teaming up with Jane, an experimental AI who is starting to develop independent thought. Shot in Northern Ireland on a modest budget, it wears its Blade Runner influence openly, but the specific question at its center, what happens the moment a manufactured mind starts noticing itself thinking, gets more screen time here than in most bigger-budget androids-with-feelings films.

Thank you for reading. We’re all beginners in something. With that in mind, your questions and feedback are always welcome and I read every single email!

-James

By the way, this is the link if you liked the content and want to share with a friend.

Some * designated product links may be affiliate or referral links. As an Amazon Associate, I earn from qualifying purchases. This helps support the newsletter at no extra cost to you and Amazon makes a tiny hair less.